Fast algorithm for reliability-guided phase unwrapping
in digital holographic microscopy
Lihong Ma,* Yong Li, Hui Wang, and Hongzhen Jin
Institute of Information Optics, Zhejiang Normal University, Jinhua 321004, China
*Corresponding author: zjnumlh@zjnu.cn
Received 27 August 2012; revised 11 November 2012; accepted 26 November 2012;
posted 27 November 2012 (Doc. ID 175033); published 20 December 2012
A fast reliability-guided phase unwrapping algorithm, using an optimized quality map and combining it
with look-up table operation, is proposed for digital holographic microscopy. First, by detecting the re-
sidues in the wrapped phase map, an intensity threshold is calculated in the normalized intensity
image and the measured region is distinguished into the reliable region and the doubtful region. An opti-
mized quality map is derived by the method in which the intensity values in the reliable region are set to 1
and those in the doubtful region remain unchanged. Then the flood fill algorithm by look-up table is
implemented with the optimized quality map to retrieve true phase map. The experimental results demon-
strate that not only does the proposed algorithm perform well, but also the speed is significantly faster than
that of the conventional flood fill algorithm using insert sorting. © 2012 Optical Society of America
OCIS codes: 090.1995, 100.5088, 100.5070, 180.6900.
1. Introduction
Digital holographic microscopy (DHM) [1,2] has be-
come a well-established instrument for the study of
microscopic samples by retrieving quantitative ampli-
tude and phase information of the wavefront trans-
mitted through or reflected from three-dimensional
(3D) objects and achieving subwavelength accuracy
along the axial direction in phase images. It has been
applied and demonstrated in many fields, such as in
the analysis and characterization of micro-electro-
mechanical systems [
3,4], the shape measurement
of micro-optical elements [
5,6], and the observation
of biological samples [
7–10]. However, in DHM, the
quantitative phase measurement is often restricted
by phase ambiguities due to the limited wavelength,
that is, only after phase extraction DHM acquires a
modulo−2π phase distribution called wrapped phase
map. F or this reason, phase unwrapping must be
carried out on the wrapped phase map to retrieve
the original continuous phase field by removing the
2π phase jumps.
Many phase unwrapping algorithms [
11–23] have
been proposed that claim to be capable of automated
unwrapping of noisy wrapped phase maps, such as
Goldstein’smethod[
12,13], the reliability-guided al-
gorithm [
14–17], Flynn’salgorithm[18], the region-
growing algorithm [
19], the fast Fourier transform/
discrete cosine transform (FFT/DCT) algorithm [
20],
and the minimum L
p
-norm algorithm [21–23]. Among
the proposed methods, the reliability-guided phase
unwrapping algorithm is one of the favorite candi-
dates for DHM, as the algorithm is surprisingly robust
in practice and can acquire an accurate unwrapped
phase. The most widely used strategy is the classical
reliability-guided flood fill algorithm [
11,14–17], in
which the sorting operation is very time-consuming.
The sorting time can be reduced by the method of con-
structing a confidence tree, in which a reliability map
is segmented into small rectangular areas with the
same reliability value, but constructing the minimum
spanning tree makes it time consuming. Anand et al.
proposed a fast flood fill algorithm by setting flood
fill thresholds step by step in which the executing
time is saved because of elimination of the tedious
sorting procedure [
24]. We have proposed a faster
flood fill algorithm called look-up table algorithm by
1559-128X/12/368800-08$15.00/0
© 2012 Optical Society of America
8800 APPLIED OPTICS / Vol. 51, No. 36 / 20 December 2012